On the Role of Preemption for Timing Metrics in Coded Multipath Communication

Recent trends in communication networks have focused on Quality of Service (QoS) requirements expressed through timing metrics such as latency or Age of Information (AoI). A possible way to achieve this is coded multipath communication: redundancy is added to a block of information through a robust packet-level code, transmitting across multiple independent channels to reduce the impact of blockages or rate fluctuation. The number of these links can grow significantly over traditional two-path schemes: in these scenarios, the optimization of the timing metrics is non-trivial, and latency and AoI might require different settings. In particular, packet preemption is often the optimal solution to optimize AoI in uncoded communication, but can significantly reduce the reliability of individual blocks. In this work, we model the multipath communication as a fork-join D/M/(K,N)/L queue, where K blocks of information are encoded into N>K redundant blocks. We derive the latency and Peak AoI (PAoI) distributions for different values of the queue size L. Our results show that preemption is not always the optimal choice, as dropping a late packet on one path might affect the reliability of the whole block, and that minimizing the PAoI leads to poor latency performance.